TgSparkHelp centerWorkspace analytics

HELP CENTER

Using workspace analytics

Review backlog, response efficiency, message volume, and account performance.

Analytics shows the current workspace's pending-reply backlog, human response efficiency, message activity, and account performance. Owners and administrators can aggregate accounts in this workspace. Ordinary members can only view accounts with explicit View grants.

Review your workspace

  1. Open Analytics and confirm the current workspace.
  2. Select dates and accounts. The default is the last 7 days; interpret dates in the workspace time zone.
  3. Start with pending and overdue replies, then review response efficiency and message trends.
  4. Select cards such as Pending or Overdue to see chat details, then open the original chat if permitted.
  5. When comparing accounts, check each account's data start, latest sync time, and known gaps.

The page refreshes every minute and can also be refreshed manually. Current backlog always reflects outstanding requests; a date filter does not hide older backlog.

Dashboard views

View What it shows
Workspace overview Pending and overdue replies, on-time response rate, typical response time, trends, and heatmaps
Response efficiency Human coverage, median and P90 first-response times, and details by response round
Messages and activity Incoming and outgoing volume, sources such as human replies, auto-replies and broadcasts, and weekday/hour distributions
Account analytics Account performance, data start, sync time, and collection gaps
Media and content Attachments, stickers, content types, observed reactions, and deletion events

Median first-response time means half of response rounds receive their first human reply within that time. P90 means roughly nine out of ten rounds do. These are different from an individual customer's waiting time.

How replies are counted

Consecutive incoming messages before the first human reply form one round, timed from the first request. In private chats, a human reply closes the current pending-response round.

In groups, only incoming messages that mention the account or reply to its messages count as requests. A human response must quote a message in that round. Ordinary group chatter and channel broadcasts are not all treated as pending requests.

Auto-replies and broadcasts count toward message volume but do not close requests waiting for a human reply. Outgoing messages with unknown sources are classified as pending confirmation; they cannot be treated as proven human replies or failed replies.

By default, a human reply within 60 minutes is on time. Owners or administrators can set 1–1,440 minutes. Unanswered requests that are not overdue are temporarily excluded from the on-time-rate denominator. Response efficiency is grouped by the date the request arrived.

Workspace analytics and ticket SLAs have separate on-time settings. Analytics calculates automatically from observed facts, while a ticket's first response also requires explicitly linking a qualifying human reply.

Understand data scope

An account enters statistics after both workspace connection and collection begin, not from its Telegram registration date. Offline backfill has limits and cannot restore every message or reaction deleted during disconnection. The dashboard retains known gaps.

The same Telegram account can be shared across workspaces. Facts about confirmed human replies may be visible across workspaces, but reports do not reveal operators from other workspaces. Workspace reports cannot simply be added together as a platform-wide total.

Chat exclusions affect analytics in the current workspace. After restoration, collection continues from a new boundary; removed history and data from the excluded period are not backfilled. Disconnecting and reconnecting does not reset this boundary.

Message-level details and observed events are retained for approximately 90 days; hourly aggregates are retained for 365 days. Historical timings older than 90 days may be estimates, and completed reply details are also subject to retention. No data means no observed data is available, not necessarily zero.

Common questions

Symptom Explanation or action
Changing the date range does not change the pending count Current backlog is not hidden by request date
An auto-reply was sent, but the chat still awaits a reply Auto-replies and broadcasts do not close requests waiting for a human
A mobile reply was not counted as a successful human reply A verifiable platform receipt is required to attribute the send; its source cannot be guessed
A group reply did not close the request Check whether it quotes a message in the corresponding round
Historical charts have gaps Check collection start, offline gaps, exclusion/restoration, and retention
The page asks for initialization Operations should run make migrate.analytics and restart HTTP; older analytics tables also require migration upgrades

For definitions, collection, and migration, see workspace analytics. Implementation: analytics dashboard.

Back to module help